Real time transfers
DIY transfer tips
Frame-by-frame transfer
Continuous motion scanners
Output Media: The third man in transfer quality!
Best possible Super 8 transfer?

Three factors determine the quality of a telecine offering: the equipment, the operator and the output media. Of the three, the operator is the wild card and perhaps the most important though least under your control. The best equipment, used by a bad operator gives bad results. Only experience (yours, or preferably other people's) will determine if the person doing the transfers cares about what they are doing. If they left a hair in the gate, chose the wrong exposure, left the image out of focus or poorly coloured, or with the frame line showing, or over cropped, etc., etc., well that's it. End of story.

So let's talk about equipment instead!

I'd like to divide the range of possible transfer set ups into three categories: real time, frame-by-frame and continuous-motion (flying spot scan).

Real Time transfers:

A 'real time' transfer involves a projector operating at or near normal projection speed and a camera recording the projector's image at the normal TV rate of 25 frames per second (PAL). The classic telecine-chain (projector, condenser lens and camera) is an instance of this. So too is the basic 'off the wall' transfer or similar methods involving rear-projection screens, transfer 'boxes', mirrors, gadgets and widgets. There are also little 'all in one' transfer units that are basically a projector with a single chip ccd pick up where the lens would be (this is a very common method of transfer out there, so be warned!).

The basic feature of the real time transfer is is a non-synchronous coupling of projector and camera. What's wrong with that? A little bit of theory will help here. Movies are generally filmed using a framing rate of 24 or 18 frames per second. That is, every second, the projector flicks through 18 or 24 distinct still pictures, employing a shutter inside the projector to block the light while the projector intermittently advances from one frame to the next. TV on the other hand uses a frame rate of 25 frames per second (at least PAL TV, the standard used in Australia, does). Every 25th of a second, the TV screen is electronically scanned in horizontal lines from top to bottom in order to make an image on the screen. Indeed, it's a little more complex than that again. Each frame of TV is itself divided into two parts known as ‘'fields'– each lasting a 50th of a second. In the first 'field' all of the odd numbered horizontal lines are scanned, then, in the second field, returning to the top of the screen the even numbered horizontal lines are scanned. These two fields, 'first' and 'second' or 'odd' and 'even' are thus 'interlaced' to make up a whole frame of image. There are two fields per frame, twenty-five frames per second, and so fifty interlaced fields per second.

The trouble for real-time telecine is fitting the 18 or 24 frames of film into these 25 frames or 50 fields of TV. They just don't fit. If left to their own devices the result can be a strobing or pulsing effect as the differing frame rates 'beat' against each other. No good. There is a solution to this, however: change the projector speed.

If the original film footage was shot at 24, the usual practice is to speed the projector up to 25 fps – the same as TV. This is only a very slight increase in speed (though it may play havoc with sound recorded with the film) and is better than pulsing. This way, the result is one frame of TV for one frame of film … almost (I'll get to the 'almost' later).

When the original film was shot at 18 fps, the best solution for real-time telecine is to slow the projector down to 16 and two-thirds frames per second. This seams an odd choice at first. On closer inspection, however, its sense is clear: 16+2/3rds goes in to 50 exactly three times. Remembering that there are 50 fields per second of TV, this means that there will be three fields corresponding to each frame of film. Again this beats the pulsing devil (but with the same 'almost' I alluded to above).

So why only 'almost'? Surely if the projector is running at 25fps and the camera is running at 25 fps there will be one TV frame per frame of film? The problem is that while the projector and the camera are running at the same speed and thus advancing at the same rate, they are nonetheless not synchronized together and nothing determines that they will start each frame at the same time as each other. If for instance the TV camera and the projector were a 50th of a second out of synch with each other, then any one frame of film will be captured by the second field of one TV frame and the first field of the next, instead of two fields from the one TV frame. A more normal case would be a random devision of each film frame amongst a spread of TV fields: perhaps part of the first field of one frame, all of the second field of that frame and part of the first field of the next frame.

There are two consequences of this. Firstly, the overall effect of a real time telecine done this non-synchronous way is a general softening of the image – some describe this nicely as a slightly dreamy effect. Secondly, and more importantly, it can make computer editing – which is done by the frame, not the field – very difficult.

Another factor to consider with real time telecine transfer is the general quality of the optical system involved. Generally a real-time transfer will involve two lenses – one on the projector and one on the camera. Often these will be zoom lenses meaning even more glass. Both of these lenses have to be perfectly focused. Further, there needs to be some intermediary element. In a classic telecine chain set up, that intermediary is a large condenser lens. In other setups there is a screen, a piece of plastic or some such other device. This third element also has its optical properties and leave its mark on the resultant footage – be it in the form of its grain or texture or other optical abberation.

A real time transfer can be quite good, especially when done using a classic telecine chain set up with an experienced operator. This is not the norm howerver. By far the majority of film to video transfer businesses use straight 'off the wall' or other low tech systems. Such systems almost always feature a pronounced 'hot spot' in the centre of the image where the projected light is brightest. Be warned. Ask to see a sample before you commit any money!

[top]

DIY OTW (off the wall!) transfer tips:

Still, I can give some tips here to those who want to do an off the wall type transfer themselves:

1) Make the projected image as small as possible. This may be determined by how small a projected image the camera can focus on, etc., but the smaller the projected image, the greater the contrast range and the colour saturation, so keep it small. The small image also helps reduce the potential hot spot from the lens.

2) Endevour to have the camera as square to the projector screen as possible.

3) It may be a good idea to put a small doughnut of cardboard right in front of the projector lens. This will act like an apeture and effectively stop down the projector lens and thus increase the depth of field of the projected image. This can help with getting edge-to-edge focus of the picture.

[top]

Frame-By-Frame Transfer:

This is a relatively new approach for film to TV transfer. Indeed it is a product of the digital age and is better described as film to digital transfer. The basic set up involves a film transporting device with an intermittent motion (just like a projector), a digital or digitized camera, a computer with appropriate capture software, and a triggering device. The film is advanced a frame at a time at a fairly slow framing rate of between say 3 and 8 fps. As each frame advances, the computer is triggered to capture and store one frame (two fields) of image. The film advances again and another frame is captured. And so it goes. After completion, the software converts what is essentially a series of still pictures of each frame of film into a digital movie file. One frame of film now corresponds to one frame of digitized data. It must be noted that due to the slow transfer speed, the sound of sound films cannot be transfered using frame-by-frame technology. For this a real-time transfer is necessary.

But what of footage shot at 18 or 24 fps? There are a few options here. Either the computer can 'blend' some frames to make up the missing 1 or 7 frames per second, it can simply duplicate 1 or 7 frames every second, or it can duplicate 2 or 14 fields every second. The last option gives the smoothest conversion, but the second option (whole frame duplication) is generally the preferred option as it avoids a version of the 'field' problem we saw with editing real-time transferred footage.

It must be remembered that not all frame by frame systems use the same equipment. The type of camera used is a significant ingredient in the overall quality, as is the type and amount of optics used, as well as the linkage of the camera's signal to the computer. One common method of frame-by-frame transfer is to use a domestic digital camera. This can yield quite adequate results, though it does mean using the domestic camera's zoom lens, as well as a condenser lens and the original projector lens, as well as being limited to a 'fire wire' (which means DV compression) linkage to the computer. It is also possible to use a broadcast quality camera with a special 'flat field' macro lens mounted on the front. This then is pointed directly at the film frame with no other optics in between. Such a camera might allow for a component video (Y/C) output to the computer or other non-compressed digital output, etc..

Frame by frame transfers are ideal for digitizing well exposed reversal (ie positive camera original) film. With minimal processing of the camera's signal it is possible to achieve a very good reproduction of the original film image (again, provided it is well exposed in the first place). They offer a cheap, very high quality transfer option. Indeed the advent of the frame by frame scanning method has been in part responsible for the renaisance of Super 8 in the digital age.

This technology, however, reaches its limit with transferring colour negative film. For this, the last and most expensive category of transfer equipment – the continuous motion flying-spot type scanner - is necessary.

[top]

Continuous-motion scanners:

Let me start by pointing out that the heading for this category is something I have made up! In the film and TV industry these machines are simply refered to by their brand names – Rank, Cintel, etc.. But this term 'continuous-motion' makes for a neat categorization in contrast to the other two above. Instead of the film advancing a frame at a time, or in real time (via the intermittent motion of a projector, its claw and shutter), the film is passed in a continuous smooth motion over a single photosensitive array - a 'flying spot', hence the conventional name 'flying spot scanner'. In the newer CCD machines, this is a single line of ccd pickups one pixel wide. With the older flying-spot machines, the pick up is a single beam from a cathode ray photo tube moving back and forth in the horizontal plane only. Division of the film back into frames is then achieved electronically by the machine. Additional 'hybrid' frames are created during transfer to compensate for the difference in frame rate between the original film and TVs 25 fps rate.

The absense of the intermitent motion of conventional film projectors means that these machines offer the steadiest possible transfer of film to video. Image steadiness in film is really a product of the precision of the mechanics involved in the intermittent transport of the film through the camera, projector and or printer gates. Most 35mm and many 16mm cameras minimise this tendency for image unsteadiness by using 'pin registration' - literally 'pining' each frame to the gate before the shutter opens and exposes the film. Unfortunately not only are super 8 film frames very small, but there are also no pin-registered super 8 cameras. This is why there is an inevitable tendency towards image unsteadiness in super 8 film (part of its 'lively' charm perhaps?). Suffice it to say, while continuous motion transfer machines they do not ordinerily process out any inherent image unsteadiness caused by the camera's intermitent motion, they nonetheless do not add any image intability of their own. In this way, continuous motion transfers will always be just that bit more stable than any of the other projector based systems discussed above.

These machines have been designed with the particularly difficult task of transferring colour negative film to video. This is difficult because of the nature of colour negative film, in particular its so called 'orange bias'. This is not a separate layer or a base colour, but part of the dye layers themselves - specifically of the green and red sensitive layers. This bias was designed in to colour negative film in order to overcome an inherent weakness in colour dye technology – namely producing an accurate photographic dye for the red or a green sensitive layers. The answer was to actually colour the red and green sensitive emulsions themselves with a colour to compensate for the error in the dye. This colouring of the emulsion remains where there was no developed dye image, and is developed out in proportion where ever there is a developed image. The result is an even colour cast across the whole developed and undeveloped parts of the image. Reversal film, of course, uses the same colour dye technology without having to have this bias. Let us remember, however, that negative film was principally designed for printing onto other film to make a positive – sometimes through several stages of printing. This colour inaccuracy would compound if the colour bias system was not used.

Continuous motion scanners then have been designed with the capacity to not only invert the colours of colour negative film, but also process out this orange bias. Further, negative film has a greatly reduced contrast range in comparison to reversal film. Today we might call this reduced contrast a form of 'compression'. This compression has to be expanded out again in the processing by these machines. As such, these machies incorporate very powerful colour correction and image processing capabilities. This capability, essential for colour negative, can also be very useful in salvaging badly under or over exposed reversal film. Further, continuous motion scanners usually also incorporate processing for reducing the appearance of scratches and dust, as well as minimizing film grain.

To harness all this power, these machines need to be operated by similarly gifted colourists. Further, the colourist must be given time to work their magic on the footage. This is why these machines are not only expensive to buy, but also expensive to operate – a good result from these machines takes time, and this the customer has to pay for. The quality of the transfer achieved will (assuming the ability of the operator) be directly proportional to the amount of suite time purchased by the customer. supervised by the client and operated by a master colourist given ample time to manipulate the footage. Conversely, a cheap transfer using one of these machines may not be worth having.


In reality only this level of machine is relevant to the 35mm/16mm film industry. The advent of Frame-by-Frame scanning has been a revolution in Super 8 film making where very often price is an important factor – especially in independent films. Further, the continuous-motion scanners require different film gates for different film gauges. These gates are expensive to buy and take time to change between and set up. Few transfer houses with these machines see Super 8 as a relevant market slice for their business. This may change.

[top]

Output Media: The third man in transfer quality!

You have to choose an output media commensurate with your needs and desires. You would be silly to spend thousands of dollars on flying spot suite time in order to output to VHS or its contemporary equivalent, the DVD. This being the digital age, your choices really come down to the question of compression and what is called 'colour sampling' or 'colour space'.

The bottom of the digital heap is the DVD. This has a compression level of about 50:1. Through using Mpeg2 compression, it is possible to fit up to 2 hours of program on one 4.7gig DVD disk. This is a great format for distributing your film to interested parties, or the video store, etc., but its no master, archival, origination or exhibition format. Further, it must be remembered that files stored on a burnt dvd are really only temporary. The disk will remain readable for a few year … maybe five or so. (Burnt disks are very different from mass produced 'stamped' disks. These may last many many years).

Next level is MiniDV. The MiniDV format is really just the consumer version of the DV format. Same tape, same compression, just a different cassette. MiniDV has been a popular origination format for home movies and some independent 'films'. It incorporates a 5:1 level of compression and what is described as 4:2:0 for PAL or 4:1:1 for NTSC colour sampling. This means that colours are 'sampled' at a quarter of the rate to the image brightness. DV compression is itself fairly costly to the original imput signal. DV compression too often results in so called 'digital artifacts' when edited. MiniDV tapes are themselves very flimsy and should not be considered an archival or exhibition medium, though they ought to remain playable long after the demise of similar age DVDs. The small and flimsy nature of MiniDV tape can sometimes create problems when playing a tape in a machine other than that it was recorded on. This is something to be aware of with transfer where this is almost always the case. Be prepared for tape failure if you are working to a tight time-line.

Digi Beata tape is really the first level in so called 'broadcast quality' digital tape. It involves very minimal compression, and colour sampling of 4:2:2 - twice that of DV. The tapes are themselves fairly expensive, as is access to playing and editing equipment. This is a good format choice for mastering, archiving and distributing your film.

Beyond digi beata are various uncompressed digital tape formats usually with 4:2:2 colour sampling. Sometimes this colour space will record each pixel of colour information using 8 bits of data, sometimes 10. This is very 'high-end' stuff and frequently beyond the reach of the super 8 film maker. Having said that, it is sometimes possible to get so called 'uncompressed', 'direct to hard drive' digital transfers of your film. Sometimes this can involve posting a portable hard-drive to the transfer facility, or else, purchasing a drive from them directly. This is a way of getting your footage in to your computer for editing without the expense of digi beta or the losses involved with lesser tape formats. As it happens, Nano Lab offers a variation on this theme for the super 8 film maker: namely transfer to uncompressed data DVD disk. This is a normal 4.7 gig DVD disk, but not recorded using DVD mpeg 2 compression. Instead, no compression is used and this is a data file only. One 50' roll of super 8 takes up 3 or 4 gig of disk space. (For more information about Nano Lab’s telecine service, see our 'telecine transfer' page). It needs to be pointed out that computers, drives and disks can be fraught with all sorts of protocol and compatibility issues. Hard drives can 'die', disks can be 'duds', file types can be wrong for your system, etc.. A DVD disk costs about a two-hundredth of the cost of a digi-beata tape and cannot be expected to be as reliable. Again, if time is short, be prepared for these sorts of issues!

So back to the operator:

With all transfers, the operator is a critical ingredient – especially with the continuous motion transfer with its powerful image processing capabilities and hence requirements. With the cheapest transfers, the machine might be set up and left unsupervised. This can be called a 'one light' transfer – that is, no adjustments are made to the transfer image in response to changes in the original film image.

Another unsupervised option, again very easy for the operator, is to use an automatic iris on the camera or pick up. This means that as the original film scene gets dark, the iris of the pick up camera compensates by opening up, and vice versa with bright scenes. With frame-by-frame transfers this can be a problem in that the camera iris can change in response to very fleeting changes in brightness of the original film – such as when a white car quickly crosses the frame. The result is a sudden darkening of the image as the car passes. Also, a 'fade to black' in the original film will be transferred as a 'fade to gray' as the iris progressively compensates for the darkening film image. Still, an automatic iris transfer can be a good option for home movies that might not otherwise warrant a closely supervised, slower and thus more expensive transfer. Indeed it might be preferable to have these constant subtle movements in image brightness than have either whole scenes too dark or too bright, or have a slower iris response made by the operator.

[top]

BEST POSSIBLE SUPER 8 TRANSFER?
So let's say you want the best. Money is no object, right? Well, take your film and cheque book to any high street broadcast telecine house with a state of the art continuous motion transfer suite and Super 8 gate (o.k.there is at least one such place in Australia - Video8 Broadcast in Sydney). Ask for their best colourist. Spend all the time necessary to grade and tweak the footage. Output to Digi Beata or better digital tape.

Can't afford that)? Go nano! - fixed price, and really quite good!
For more details of Nano Lab's transfer system, please see our 'telecine transfer' page. After reading all this, you are now ready for it!